AI-powered fraud detection in banking : innovations, challenges and preventive strategies

Abstract

As artificial intelligence (AI) becomes integrated in the banking industry, it enhances efficiency and introduces vulnerabilities while AI-driven frauds have emerged as a significant threat. This study explores the risks of AI-enabled frauds and strategies for prevention. Through literature review, secondary data analysis, and examples, this thesis identifies key fraud types, including generative AI’s role in creating deepfakes for signatures, videos, and voice impersonations, which cause financial losses and undermine trust.<br>The thesis identifies cutting-edge countermeasures to these risks, including shared large language models (LLMs), automation tools, liveness testing, compliance controls, and machine learning-based anti-fraud technology. These methods strengthen data security and enhance fraud detection. The results highlight the critical necessity for flexible tactics to counteract changing fraud schemes and provide helpful advice for bolstering financial defenses and guaranteeing the stability of banking institutions in a time of swift technological advancement.